Disparities in mortality from prostate and colorectal cancers between African Americans and Caucasians in the Chicago area are either higher than the national average or increasing. Although the factors responsible are not known, preliminary data from hospital-based cohorts suggest they involve persistent inequalities in advanced-stage disease at diagnosis, socioeconomic disadvantage, and the accessibility and pattern of cancer care. The objective of this proposal is to conduct population-based analyses of clinical, socioeconomic, and healthcare-related determinants of the disparities in mortality from prostate and colorectal cancer that exists between African Americans and Caucasians in the Chicago area. SPECIFIC AIMS: 1) Create a database of incident prostate and colorectal cancers diagnosed in the Chicago area between 1995 and 2008 that integrates data on case demographics, tumor characteristics, and patterns of initial cancer care with data on the administrative characteristics of the diagnosing and treating facility, spatial and non-spatial measures of access to healthcare at the census tract level, and cause-specific mortality. Data source include the Illinois State Cancer Registry, 1990 and 2000 U.S. Decennial Census, American Medical Association, and the National Death Index. 2) Conduct retrospective cohort studies of prostate cancer and colorectal cancer to evaluate the extent to which racial disparities in cancer-specific mortality are due racial differences in stage at diagnosis, patterns care, census tract-level socioeconomic disadvantage ("non-spatial access"), and geographical access to healthcare providers ("spatial access"). The main hypotheses to be tested is that racial differences in stage, treatment, socialeconomic deprivation, and geograpohical access to healthcare at the census tract level medicate the disparities in cancer-specific mortality. 3) Use the census tract-level data on healthcare access to develop models to predict which areas within Chicago area are future "hotspots" for advanced-stage prostate and colorectal cancer. Census tracts with a higher-than-expected prevalence of advanced-stage prostate and colorectal cancer ("hotspots") will be identified using spatial scan statistics. Logistic regression models will identify spatial and non-spatial characteristics at the census tract level that associate with an excessive risk of advance-stage tumors.